1. Field of the Invention
The present invention relates to a crystal growth method, a crystal growth apparatus, a group-III nitride crystal, and a group-III nitride semiconductor device. In particular, the present invention relates to a crystal growth method and a crystal growth apparatus for a group-III nitride crystal, the group-III nitride crystal, and a group-III nitride semiconductor device employing the group-III nitride crystal applicable to a blue-violet light source, an ultraviolet light source (LED or LD) for an optical disk drive, a blue-violet light source for electro-photography printing system, and so forth.
2. Description of the Related Art
Now, a InGaAlN-family (group-III nitride) device used as violet through blue through green light sources is produced by a crystal growth process employing an MO-CVD method (organic metal chemical vapor phase growth method), an MBE method (molecular beam crystal growth method), etc. on a sapphire or SiC substrate in most cases. In using sapphire or SiC as a substrate, crystal defect caused due to a large expansivity difference and/or lattice constant difference from a group-III nitride may occur frequently. By this reason, there is a problem that the device characteristic may become worth, it may be difficult to lengthen the life of the light-emission device, or the electric power consumption may become larger.
Furthermore, since a sapphire substrate has an insulating property, drawing of an electrode from the substrate like in another conventional light-emission device is impossible, and therefore, drawing the electrode from the nitride semiconductor surface on which crystal was grown is needed. Consequently, the device area may have to be enlarged, and, thereby, the costs may increase. Moreover, chip separation by cleavage is difficult for a group-III nitride semiconductor device produced on a sapphire substrate, and it is not easy to obtain a resonator end surface needed for a laser diode (LD) by cleavage, either. By this reason, a resonator end surface formation according to dry etching, or, after grinding a sapphire substrate to the thickness of 100 micrometers or less, a resonator end surface formation made in a way near cleavage, should be performed. Also in such a case, it is impossible to perform formation of a resonator end surface and chip separation easily by a single process like for another conventional LD, and, also, complication in process, and, thereby, cost increase may occur.
In order to solve these problems, it has been proposed to reduce the crystal defects by employing a selective lateral growth method and/or another technique for forming a group-III nitride semiconductor film on a sapphire substrate.
For example, a document ‘Japanese Journal of Applied Physics, Vol. 36 (1967), Part 2, No. 12A, pages L1568-1571’ (referred to as a first prior art, hereinafter) discloses a laser diode (LD) shown in FIG. 1. This configuration is produced as follows: After growing up a GaN low-temperature buffer layer 2 and a GaN layer 3, one by one, on a sapphire substrate 1 by an MO-VPE (organometallic vapor phase epitaxy) apparatus, a SiO2 mask 4 for selective growth is formed. This SiO2 mask 4 is formed through photo lithography and etching process, after depositing a SiO2 film by another CVD (chemistry vapor phase deposition) apparatus. Next, on this SiO2 mask 4, again, a GaN film 3′ is grown up to a thickness of 20 micrometers by the MO-VPE apparatus, and, thereby, GaN grows laterally selectively, and, as a result, the crystal defects are reduced as compared with the case where the selective lateral growth is not applied. Furthermore, prolonging of the crystal defect toward an activity layer 6 is prevented by provision of a modulation doped strained-layer superlattice layer (MD-SLS) 5 formed thereon. Consequently, as compared with the case where the selective lateral growth and modulation doped strained-layer superlattice layer are not used, it becomes possible to lengthen the device life.
In the case of this first prior art, although it becomes possible to reduce the crystal defects as compared with the case where the selective lateral growth of a GaN film is not carried out on a sapphire substrate, the above-mentioned problems concerning the insulating property and cleavage by using a sapphire substrate still remain. Furthermore, as the SiO2 mask formation process is added, the crystal growth by the MO-VPE apparatus is applied twice, and, thereby, a problem that a process is complicated newly arises.
As another method, for example, a document ‘Applied Physics Letters, Vol. 73, No. 6, pages 832-834 (1998)’ (referred to as a second prior art, hereinafter) discloses application of a GaN thick film substrate. By this second prior art, a GaN substrate is produced, by growing up a 200-micrometer GaN thick film by an H-VPE (hydride vapor phase growth) apparatus after 20-micrometer selective lateral growth according to the above-mentioned first prior art, and, then, grinding the GaN substrate thus having grown to be the thick film from the side of the sapphire substrate so that it may have the thickness of 150 micrometers. Then, the MO-VPE apparatus is used on this GaN substrate, crystal growth processes required for an LD device are performed, one by one, and, thus, the LD device is produced. Consequently, it becomes possible to solve the above-mentioned problems concerning the insulating property and cleavage by using the sapphire substrate in addition to solving the problem concerning the crystal defects.
A similar method is disclosed by Japanese Laid-Open Patent Application No. 11-4048 and FIG. 2 shows a device configuration according to this method.
However, the process is more complicated in the second prior art, and, requires higher costs, in comparison to the first prior art. Moreover, in growing the no less than 200 micrometer GaN thick film by the method of the second prior art, a stress occurring due to a lattice constant difference and an expansivity difference from the sapphire of the substrate becomes larger, and a problem that the curvature and the crack of the substrate may newly occur. Moreover, even by performing such a complicated process, the crystal defect density can be reduced to only on the order of 106/cm2. Thus, it is not possible to obtain a practical semiconductor device.
In order to avoid this problem, setting to 1 mm or more thickness of an original substrate (sapphire and spinel are the most desirable materials as the substrate) from which a thick film grows is proposed by Japanese Laid-Open Patent Application No. 10-256662. According thereto, no curvature nor crack arise in the substrate even when the GaN film grows in 200 micrometers of thickness by applying this substrate having the thickness of 1 mm or more. However, a substrate thick in this way has a high cost of the substrate itself, and it is necessary to spend much time on polish thereof, and leads to the further cost rise of the polish process. That is, as compared with the case where a thin substrate is used, the cost becomes higher by using the thick substrate. Moreover, although no curvature nor crack arise in the substrate after growing up the thick GaN film in using the thick substrate, curvature and/or crack may occur as stress relief occurs during the process of polish. By this reason, even when the thick substrate is used, the GaN substrate having a high crystal quality and having such a large area that it can be practically used for an ordinary semiconductor device manufacturing process cannot be easily produced.
A document ‘Journal of Crystal Growth, Vol. 189/190, pages. 153-158 (1998)’ (referred to as a third prior art, hereinafter) discloses that a bulk crystal of GaN is made grow up, and it is used as a homoepitaxial substrate. According to this technique, under the high temperature in the range between 1400 and 1700° C., and under the very high nitrogen pressure of 10 kilobars, crystal growth of the GaN is performed from a Ga liquid. In this case, it becomes possible to grow up a group-III nitride semiconductor film required for a device by using this GaN substrate. Therefore, it is possible to provide the GaN substrate without needing a process complicate like in the above-described first and second prior arts.
However, by this third prior art, crystal growth in high temperature and high pressure is needed, and, thus, there is a problem that a reaction vessel which can resist these conditions should be very expensive. In addition, even when such a growth method is employed, the size of the crystal obtained has the problem of being too small, i.e., at most on the order of 1 cm, and, thus, it is too small to put it in practical use of semiconductor device manufacture.
The GaN crystal growth method using Na which is an alkaline metal as a flux is proposed by a document ‘Chemistry of Materials, Vol. 9 (1977), pages 413-416’ (referred to as a fourth prior art, hereinafter) as a technique of solving the problem of GaN crystal growth in the above-mentioned high temperature and high pressure. According to this technique, sealing sodium azide (NaN3) and Ga metal used as a flux and a material into a reaction vessel made from stainless steel (vessel inner dimension: diameter=7.5 mm and length=100 mm) in nitrogen atmosphere, and the reaction vessel is maintained in the temperature in the range between 600 and 800° C. for 24 to 100 hours to grow up a GaN crystal. In the case of this fourth prior art, crystal growth at the comparatively low temperature in the range between 600 and 800° C. can be achieved, and, also, the require pressure inside the vessel should be only on the order of 100 kg/cm2, which is comparatively lower than the case of the third prior art. However, in this fourth prior art, the size of the crystal obtained is small as less than 1 mm which is too small to be put into practical use in semiconductor device manufacture, like in the case of the third prior art.
Therefore, the applicant of the present application has proposed a method of enlarging a group-III nitride crystal. However, in the method, nucleus generation initiates of the crystal growth is natural nucleus generation, and, thus, a large number of nucleus are undesirably generated. In order to control this nucleus generation, the applicant has proposed to utilize a seed crystal in the U.S. patent application Ser. No. 09/590,063, filed on Jun. 8, 2000, by Seiji Sarayama et al. (the entire contents of which are hereby incorporated by reference). However, there is a problem that a required crystal growth apparatus becomes complicated. Therefore, it has been demanded to realize a method for effectively controlling nucleus generation, while achieving a simple apparatus configuration of a conventional flux method, in order to solve this problem.
Further, Japanese Laid-Open Patent Application No. 2000-327495 discloses a fifth prior art combining the above-mentioned fourth prior art and an epitaxial method utilizing a substrate. In this method, a substrate on which GaN or AlN is grown previously is used, and, thereon, a GaN film according to the fourth prior art is grown. However, in this method, as it is basically the epitaxial method, the problem of crystal defects occurring in the above-mentioned first and second prior arts cannot be solved. Further, as the GaN film or AlN film should be grown on the substrate previously, the process becomes complicated, and, thereby, the costs increase.
Furthermore, recently, Japanese Laid-Open Patent Applications Nos. 2000-12900 and 2000-22212 disclose a sixth prior art in which a GaAs substrate is used and a GaN thick-film substrate is produced. In this method, a GaN film having a thickness in a range between 70 μm and 1 mm is selectively grown on a GaAs substrate by using an SiO2 film or SiN film as a mask as in the above-mentioned first prior art, as shown in FIGS. 3A through 3C. The crystal growth there is performed by the H-VPE apparatus. Then, the GaAs substrate is etched and thus removed by using aqua regia. Thus, the GaN self-standing substrate is produced, as shown in FIG. 3D. By using this GaN-self standing substrate, a GaN crystal having a thickness of several tens of millimeters is grown by vapor phase epitaxy by the H-VPE apparatus again, as shown in FIG. 4A. Then, this GaN crystal of several tens millimeters is cut into wafer shapes by a slicer, as shown in FIG. 4B. Thus, GaN wafers are produced, as shown in FIG. 4C.
According to this sixth prior art, the GaN self-standing substrate can be obtained, and, also, the GaN crystal having the thickness of several tens of millimeters can be obtained. However, this method has the following problems:                {circle around (1)} As the SiN film or SiO2 film is used as a mask for selective growth, the manufacturing process becomes complicated, and, thus, the costs increase;        {circle around (2)} When the GaN crystal having the thickness of several tens of millimeters is grown by the H-VPE apparatus, GaN crystals (in monocrystal or polycrystal) or amorphous GaN having a similar thickness adhere to the inner wall of the reaction vessel. Accordingly, the productivity/yield is degraded thereby.        {circle around (3)} As the GaAs substrate is etched and removed every time of the crystal growth as a sacrifice substrate, the costs increase thereby.        {circle around (4)} With regard to the crystal quality, problems of lattice mismatch due to crystal growth on a different-substance substrate, and a high defect density due to difference in expansivity remain.        